Fig 1.
BMPR2-deficient ECs gain canonical TGFβ-SMAD2/3 and lateral TGFβ-SMAD1/5 responsiveness.
(A) EC homeostasis is controlled by balanced TGFβ and BMP signaling. Loss of BMPR2 (black) leads to unbalanced TGFβ/BMP signaling. BMP6/2 activate ECs via BMPR2-ALK2 and BMPR2-ALK3 signaling complexes, respectively, to induce angiogenesis. Receptor complexes of TβR2 with either ALK5 alone (blue arrow), with ALK5 and ALK1, or with ALK5 and ALK2/3 (blue lines) have been described. TGFβ-induced activation of SMAD1/5/8 via such a complex is termed “lateral TGFβ signaling.” BMP9 acts via ALK1-SMAD1/5/8 signaling as a vascular quiescence factor important to maintaining vessel integrity. (B) Immunoblot using an antibody specific to pSMAD1/5. Cells were treated with BMP6 (3 nM) or BMP9 (0.3 nM) for 15 min. Densitometric quantification (right) of pSMAD1/5 relative to GAPDH levels expressed as AUs (n = 7 independent experiments). (C) Immunoblot (left) using antibodies specific to pSMAD1/5 or pSMAD2. Cells were treated with TGFβ (200 pM) for 15 min. Quantification of pSMAD1/5 and pSMAD2 signal intensity relative to GAPDH levels (right) expressed as AUs (n = 6 independent experiments). (D) Immunoblot showing dose responses for TGFβ (50 pM, 200 pM, 600 pM), BMP9 (50 pM, 150 pM, 300 pM), or BMP6 (1.5 nM, 3 nM, 10 nM) after 15 min of stimulation. The ratios of signal intensities are shown below each panel. (E) Total protein levels under steady-state growth conditions of indicated SMADs. Quantification of tSMAD1–3 signal intensity relative to GAPDH levels (right) expressed as AUs (n = 8–12 independent experiments). (F) qRT-PCR (6 h of starvation and 3 h [for ID3] or 24 h [for CTGF] stimulation) with indicated ligands. Values are expressed as F.I. (n = 3 independent experiments). In all panels, the data are shown as mean + SD relative to BMPR2wt. Statistical significance relative to BMPR2wt was calculated using Kruskal-Wallis test with post hoc Dunn test for densitometric quantifications and two-way ANOVA and Bonferroni post hoc test for RT data. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. See also S2 Fig and S1 Data for underlying data. ALK2, ; AU, arbitrary unit; BMP, bone morphogenetic protein; BMPR2, BMP type-2 receptor; EC, endothelial cell; F.I., fold induction; n.s., not significant; pSMAD, ; qRT-PCR, quantitative real-time PCR; RT, real time; SMAD, Suppressor of Mothers against Decapentaplegic; TβR2, TGFβ receptor 2; TGFβ, transforming growth factor-beta; tSMAD, total SMAD.
Fig 2.
BMPR2-deficient ECs signal through heteromers comprising BMP and TGFβ receptors as indicated by the formation of mixed SMAD complexes.
(A) Scheme depicting the targeting conditions for TGFβ-induced SMAD1/5 and SMAD2/3 signaling, i.e., inhibition of TβR2-ALK5 and heteromeric TβR2-ALK5-ALK1/ALK2 complexes by selective small molecules K02288 against ALK2/ALK1 or SB-431542 against ALK5, or siRNA targeting TβR2. (B) Immunoblot using antibodies specific to pSMAD1/5 or pSMAD2 showing responses upon 15 min TGFβ stimulation (200 pM) with a 1-h pre-exposure to K02288 (0.5 μM), SB-431542 (5 μM), or DMSO. (C) Effects of TβR2 knock-down by specific siRNA compared to a scrambled control. (D) Densitometric quantification of pSMAD1/5 relative to GAPDH levels expressed as AUs (n = 4 independent experiments). Note the significant reduction in lateral TGFβ signaling on the level of pSMAD1/5 phosphorylation (right) when TβR2 levels are reduced. (E) Epifluorescence images of PLA (left) showing complexes of SMAD1 (S1) with SMAD2 (S2) in indicated cell clones upon TGFβ stimulation (200 pM) for 15 min. PLA signals are pseudo-colored greyscale and inverted (upper). Scale bar, 10 μm. Quantification of SMAD1-SMAD2 PLA signals (right) in TGFβ-stimulated cells with the number of nuclear, cytosolic, and overall PLA foci shown. Data are presented as mean ± SD (n = 9 frames, 20–30 cells each). Statistical significance relative to BMPR2wt was calculated using two-tailed Mann-Whitney test for densitometric quantification and one-way ANOVA and Bonferroni post hoc test for PLA data. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. See also S3 Fig and S2 Data for underlying data. ALK5, activin receptor-like kinase 5; AU, arbitrary unit; BMP, bone morphogenetic protein; BMPR2, BMP type-2 receptor; EC, endothelial cell; PLA, proximity ligation assay; pSMAD, phosphorylated SMAD; siRNA, small interfering RNA; SMAD, suppressor of mothers against decapentaplegic; TβR2, TGFβ type-2 receptor; TGFβ, transforming growth factor-beta; tSMAD, total SMAD.
Fig 3.
BMPR2 deficiency alters integrin expression.
(A) Immunoblot using antibodies specific to pSMAD1/5 or pSMAD2 showing elevated pSMAD levels in indicated cell types cultivated in confluent monolayers for 3 days under steady-state conditions. (B–E) RNA-Seq analysis of WT and BMPR2-deficient ECs under steady-state conditions (n = 3 independent replicates). (B) Number of genes significantly differentially regulated in BMPR2-deficient ECs in comparison to WT ECs and their relative proportion. The majority of genes are similarly altered in both BMPR2-deficient cells lines (829 = down; 833 = up). (C) Significantly enriched GO terms of shared deregulated genes. The fold enrichment of up-regulated (red bars) and down-regulated (blue bars) GO terms is shown. Notably, GO terms associated with cell-to-cell and cell-to-substrate connectivity were both up- and down-regulated, suggesting an alteration of the cellular mechanics in absence of BMPR2. (D) Hierarchical clustering of differentially expressed genes associated with the GO term “cell junction” (GO: 0030054). Heatmap color coding shows z-score of differentially regulated genes (red = high; blue = low). (E) Relative expression of integrins under steady-state conditions shown with RPKM values. Note ITGB1 is significantly elevated and the most abundant integrin in BMPR2-deficient ECs. Statistical significance relative to BMPR2wt was calculated using one-way ANOVA and Bonferroni post hoc test. *P < 0.05, **P < 0.01, ***P < 0.001. (F) IGV browser displays over the ITGB1 loci showing SMAD1/5 ChIP-Seq track of HUVECs treated with BMP9 [53] and pSMAD1/5 ChIP-Seq track of MDA-MB-231 cells treated with TGFβ1 [41]. ChIP-Seq data were retrieved from GEO (GSM684747, GSM2429820). (G) SMAD1 occupancy at the ITGB1 TSS or the 22-kb regions were validated by ChIP-qPCR in steady-state conditions. IPs are a representative experiment of two and ChIP-qPCR was performed in triplicates shown with mean + SD. See also S4 Fig and S3 Data for underlying data. BMP, bone morphogenetic protein; BMPR2, BMP type-2 receptor; BP, biological process; CC, cellular compartment; ChIP, chromatin immunoprecipitation; EC, endothelial cell; GEO, Gene Expression Omnibus; GO, Gene Ontology; HUVEC, human umbilical vein endothelial cell; IgG, immunoglobulin G; IP, immunoprecipitation; pSMAD, phosphorylated SMAD; qPCR, quantitative PCR; RNA-Seq, RNA sequencing; RPKM, Reads per kilobase per million mapped reads; TGFβ, transforming growth factor-beta; TSS, transcription start site; WT, wild-type; CC FAT, cellular compartment; BP FAT, biological process.
Fig 4.
BMPR2-deficient ECs show increased activation of β1-integrin-ILK mechano-complexes, re-localization of these complexes to cell junctions, and junctional stiffening.
(A) Immunoblot showing total β1-integrin (tITGB1) levels and phosphorylation at Ser785 and Threonine 783 (left) under steady-state conditions and densitometric quantification of total or phosphorylated β1-integrin levels in indicated cell clones (right). Data are presented as mean + SD (n = 3–4 independent experiments). (B) PLA between active β1-integrin (ITGB1pS785) and ILK under steady-state growth conditions. (C) Single confocal z-planes of BMPR2wt (basal, left), BMPR2ΔE2 (medial, middle), and BMPR2KO (medial, right) depicting the localization of ILK (green) or paxillin (red). Insets show zoomed-in regions. (D) Immunoblots showing protein levels of VE-Cadherin and phosphorylated VE-Cadherin in indicated ECs under steady-state conditions. (E) Single confocal z-planes of immunocytochemical staining using antibodies specific for VE-Cadherin (green) and PECAM-1 (red). Insets show zoomed-in regions. (F) Immunoblots showing levels of phosphorylated cofilin (pSer-3), pMLC (pSer-19), and total MLC in indicated ECs under steady-state culture conditions. (G) Cartoon (left) depicting the principle of CFS (particle diameter 23 μm). Elastic modulus derived from CFS of living cells (middle). Representative indentation (force versus distance) curves for the different cell lines and the hard control surface (mica; right). Data are shown as mean + SD (n = 30). (H) Principle of QI using a sharp cantilever tip (diameter < 20 nm; left). Representative QI scans of fixed cells, focusing on CCC sites (middle). Height profiles of nucleus-to-cell junction height differences (black arrows) are indicated. Given values are expressed as the mean ± SD. Height profile measurements were taken from n ≥ 18 cells; scale bars, 10 μm; Statistical significance relative to BMPR2wt was calculated using Kruskal-Wallis test with post hoc Dunn test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. See also S5 Fig and S4 Data for underlying data. BMP, bone morphogenetic protein; BMPR2, BMP type-2 receptor; CCC, cell-to-cell contact; CFS, colloidal force spectroscopy; EC, endothelial cell; ILK, integrin-linked kinase; MLC, myosin light chain; PECAM-1, platelet endothelial cell adhesion molecule-1; PLA, proximity ligation assay; pMLC, phosphorylated MLC; QI, quantitative imaging; pS785, phosphorylated Serine 785; pT783, phosphorylated Threonine 783; tITGB1, total integrin beta-1; VE, vascular endothelial.
Fig 5.
BMPR2-deficient ECs display altered ECM expression profile, spread on and remodel RGD-containing ECM, and exert high traction forces dependent on β1-integrin activity.
(A) RNA-Seq analysis of WT and BMPR2-deficient ECs under steady-state conditions (n = 3 independent replicates). Hierarchical clustering of differentially expressed genes associated with the GSEA gene set “NABA_CORE_MATRISOME” (M5884). Heatmap color coding shows z-score of differentially regulated genes (red = high; blue = low). (B) Cell adhesion on dishes (TC plastic) coated with ECM proteins (all 5 μg/cm2). Spreading area is expressed as AUs. Data are presented as mean + SD (two-way ANOVA with post hoc Bonferroni relative to BMPR2wt, n = 3 independent experiments). See also S6 Fig for representative images of cells. (C) ECIS measurement at 64 kHz over 30 h monitoring cell adhesion and spreading on gelatin-coated surfaces showing decay of mean capacitance (nF) ± SEM for WT and BMPR2-deficient cells upon seeding in basal growth media and when cells were additionally pre-incubated with nonspecific antibody isotype control (IgG) or β1-integrin blocking antibody against the ecto-domain (slash-dotted line). Two-hour time point is depicted as mean capacitance ± SD (right), showing that interference with β1-integrin function by blocking antibodies resembles IgG control for WT cells; two-way ANOVA with post hoc Bonferroni relative to IgG BMPR2wt was performed for n = 2–4 independent experiments. (D) ECIS adhesion assay upon 3 d SMKI treatment targeting ALK5 or ALK1/2 with SB-431542 and K02288. Plot shows mean capacitance (nF) ± SD, 2 h after seeding normalized to 0 h. Red line represents mean capacitance of DMSO-treated BMPR2wt cells. Note that BMPR2-deficient cells have lost their increased adhesion and spreading potential compared to BMPR2wt when cells were subjected to ALK5, ALK1/2, and double SMKI treatment (n ≥ 3 independent experiments). Statistical significance was calculated between DMSO-treated WT and BMPR2-deficient cells using an unpaired two-tailed Student t test. (E) Collagen lattice contractility assay showing contraction of collagen I lattice co-cultured with indicated EC clones in the absence (DMSO) or presence of ROCK inhibitor Y-27632 (10 μM) (upper). Quantification of lattice diameter constriction (as compared to initial diameter at 1 h) expressed as percent of initial diameter after 24 h of incubation (lower). Data are presented as mean + SD (two-way ANOVA with post hoc Bonferroni relative to DMSO BMPR2wt, n = 3 independent experiments). (F) Immunocytochemical staining of indicated single cells growing on FN-coated circular micropatterns for 24 h showing the relative localization of ILK (green) and remodeled FN (red) (upper). Representative epifluorescence images for different diameter circular micropatterns cultured with single BMPR2wt (i) or BMPR2ΔE2 (ii) cells, depicting underlying FN (red) coating after 24 h (lower). Scale bars, 10 μm. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. See also S6 Fig and S5 Data for underlying data. ALK5, activin receptor-like kinase 5; AU, arbitrary unit; BMP, bone morphogenetic protein; BMPR2, BMP type-2 receptor; EC, endothelial cell; ECIS, electric cell-substrate impedance sensing; ECM, extracellular matrix; FN, fibronectin; GSEA, Gene Set Enrichment Analysis; IgG, immunoglobulin G; ILK, integrin-linked kinase; n.s., not significant; RNA-Seq, RNA sequencing; ROCK, Rho-associated kinase; SMKI, small-molecule kinase inhibitor; TC, tissue culture; WT, wild type.
Fig 6.
BMPR2-deficient ECs remodel Fibronectin and Fibrillin particularly at β1-integrin-ILK–rich cell junctions.
(A) Epifluorescence images showing indicated cell clones cultured for 3 d in the presence of FNrho (20 μg/ml) (white, pseudo-color) and counterstained for cell membrane (DiO, green). Figure enlargements are indicated by white frame (lower). Enlargement showing relative localization of cell boundaries (green) and FN fibers (white) indicated by white arrowhead. Scale bar, 20 μm. (B) Immunoblot using an antibody against FBN1 on conditioned supernatants of indicated cell clones (upper) together with Ponceau-S staining (lower). Cells were cultured under steady-state growth conditions, and equal volumes of supernatants were harvested after 2, 7, and 10 d and concentrated. (C) Epifluorescence pictures/images of indicated cell clones seeded on circular micropatterns coated with FN (red). Immunocytochemical staining shows FBN1 (white pseudo-color) after 24 h of culture deposited on FN-coated micropatterns. Sites of active co-remodeling are indicated by white arrowheads. (D) Quantification of FBN deposits on 15 mm coverslips after decellularization (upper) using a fluorescence laser scanner (Cy3 emission). The quantified coverslip area is indicated by slash-dotted circle (lower). Data are presented as mean + SD (one-way ANOVA with post hoc Bonferroni relative to BMPR2wt, n = 3). (E) Re-constructed confocal z-stacks (volume rendering) of FBN1 immunostaining depicted by topographical color-coding, indicating relative z-position of fluorescent signal (approximately 0–4 μm red-yellow/basal; approximately 4–7 μm green/medial; approximately 8–11 μm blue/apical). Note the large number of medially located signals. (F) Maximum projections of confocal z-stacks showing relative localization of β1-integrin (ITGB1, green) and FBN1 (red). Co-localization is indicated by white arrowheads (right). See S7B Fig for side-view projection. (G) Epifluorescence images showing relative ILK (green) and FBN1 (red) localization upon control (scrambled) si-RNA transfection (upper) or FBN1 knock-down (lower). Scale bars, 10 μm (panels A, C, F, and G) and 50 μm (panel E). See also S7 Fig and S6 Data for underlying data. BMP, bone morphogenetic protein; BMPR2, BMP type-2 receptor; Cy3, cyanine dye 3; DiO, green fluorescence emission lipophilic carbocyanine dye; EC, endothelial cell; FBN, fibrillin; FN, fibronectin; FNrho, rhodamine-labeled FN; ILK, integrin-linked kinase.
Fig 7.
Ectopic FBN1 deposits, actomyosin contractility, and loss of endothelial character found in inner luminal PAs from IPAH and HPAH donors with low BMPR2 expression.
(A) qRT-PCRs of whole lung tissue from IPAH and HPAH donors (n = 19) analyzed for their BMPR2 transcript levels. Red line indicates mean BMPR2 transcript levels in control tissue (n = 9). (B) Immunohistochemical stainings of control and HPAH tissue sections showing large (ii) and small (ii) PAs at the level of the terminal bronchioles; 10-μm-thick sections were stained for FBN1 (red) and DAPI (blue). Autofluorescence of elastic membranes (collagen and elastin) at approximately 520 nm emission (green) was used to identify relative locations of iem and eem. Note hypertrophy of intima and thickened tm in HPAH PAs when compared to control. Ectopic FBN1 deposits are found to exceed the iem toward the lumen and to add up to intimal hypertrophy. Concomitantly, few FBN1 deposits are found in medial, while larger deposits are found in adventitial regions. (C) Quantification of luminal FBN1 deposits in PAs from regions beyond the border indicating the iem; ≥10 different PAs originating from up to 3 different donors were compared (control n = 3, IPAH n = 3, HPAH n = 1). (D) Representative PAs of controls, IPAH, and HPAH donors were stained for FBN1, collagen, and elastin at approximately 520 nm emission (green) and DAPI (blue) (i). Higher magnification of the area surrounding the iem (ii) shows clear restriction of FBN1 to sub-EC layer (e.g., basal lamina) in controls, while FBN1 deposits are also found to extend into the lumen of PAs from IPAH (middle) and HPAH (right) donors. Note the fragmented structure of FBN1 deposits in media of IPAH plexiform lesions. In contrast, low amounts of medial FBN1 deposits versus pronounced eem, adventitial and intimal localization is found in HPAH PAs (asterisks, arrowheads). (E) Sections were stained for pMLC (red) indicating actomyosin-dependent contractility. In controls, pMLC signals are found confined in close proximity to the iem, whereas in both IPAH and HPAH, pMLC is also found exceeding the iem toward the luminal side. (F) PECAM-1 staining (red) was used to show localization of ECs in PAs from controls and IPAH/HPAH. In controls, ECs locate with distinct inner luminal distribution and connectivity in close proximity to the iem. In plexiform lesions of IPAH and HPAH, few PECAM-1–positive cells could be identified in far distance to the iem indicating loss of endothelial character. Scale bar represents 100 μM (panel B) or 50 μm (panels D–F). See also S8 Fig and S7 Data for underlying data. ad, adventitial; BMP, bone morphogenetic protein; BMPR2, BMP type-2 receptor; EC, endothelial cell; eem, external elastic membrane; FBN1, fibrillin-1; HPAH, heritable PAH; iem, internal elastic membrane; IPAH, idiopathic PAH; it, intima; lu, lumen; PA, pulmonary artery; PECAM-1, platelet endothelial cell adhesion molecule-1; pMLC, phosphorylated myosin light chain; qRT-PCR, quantitative real-time PCR; tm tunica media.
Fig 8.
The mechano-adaptation of BMPR2-deficient ECs leads to increased retrieval of TGFβ from extracellular latency depots.
(A) Cartoon depicting the structure of the LLC tethering the SLC via LTBP to fibrillin. TGFβ interacts with LAP exposing RGD sequence. Disulfide bonds are depicted in orange. (B) Total cell lysates including cell extract and ECM deposits analyzed using antibodies specific to LAP of TGFβ1 (upper panel) and LTBP-1 (middle panel) as well as mature TGFβ (lower panel). Different variants of LAP complexes are identified under non–fully-reducing conditions including LAP-only monomers (approximately 35 kDa), LAP as part of the monomeric SLC (approximately 50 kDa), and LAP-only dimers (approximately 70 kDa), of which the mature ligand has been cleaved off and retrieved by cells. The upper bands of these individual forms reflect PTMs as can be predicted in silico. (C) Single confocal z-planes (BMPR2wt = basal; BMPR2ΔE2 = medial) showing stainings of FBN1 (red) and LAP (green) in intact cells (i). Figure enlargements are indicated by white frame (ii). Exemplary co-localization is indicated by white arrowheads (lower). Line scans (blue line; ii) for fluorescence intensities of FBN1 (red) and LAP (green) are indicated (iii). (D) Maximum projections of confocal stacks of decellularized ECM showing staining of FBN1 (red) and LAP (green). (i) Lack of DAPI staining indicates successful decellularization (ii) for indicated cell clones. Figure enlargements are indicated by white frame (ii). Exemplary co-localization shown by white arrowheads (lower). Line scans (ii) for fluorescence intensities of FBN1 (red) and LAP (green) are indicated (iii). (E) Cartoon describing bioassay using stable CAGA12-Luc TGFβ reporter cells and acidification of conditioned SNs. (F) TGFβ bioassay showing firefly luciferase values relative to renilla luciferase values (in RLU) for conditioned SNs of WT and BMPR2-deficient cells that were added to reporter cells either untreated (left), or when same conditioned SNs were treated with acid to activate latent complexes of TGFβ (middle) and expressed as relative ratio of acidified to nonacidified showing increase in latent TGFβ in conditioned SN of BMPR2-deficient cells (right). (G) Scheme depicting the strategy of decellularization of BMPR2-mutant ECM and re-cellularization to report on ECM-bound TGFβ by use of a bioassay. Cells were left to produce ECM followed by specific decellularization method. (H) Decellularized ECM was re-cellularized with HEK293T cells transiently transfected with a TGFβ reporter construct (CAGA12-Luc) (Kruskal-Wallis test with post hoc Dunn test, n = 12) (left) or when reporter cells were concomitantly overexpressing ITGB1 or when cells were additionally cultured in presence of either DMSO or ROCK inhibitor Y-27632 (20 μM; right) (two-way ANOVA with post hoc Bonferroni relative to DMSO, n = 3–4). In all panels, data are shown as mean + SD; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. See also S7E Fig and S8 Data for underlying data. BMP, bone morphogenetic protein; BMPR2, BMP type-2 receptor; EC, endothelial cell; ECM, extracellular matrix; FBN1, fibrillin-1; LAP, latency-associated peptide; LLC, large latency complex; LTBP, latent transforming growth factor beta-binding protein; PTM, post-translational modification; RLU, relative light unit; ROCK, Rho-associated kinase; SLC, small latency complex; SN, supernatant; TGFβ, transforming growth factor-beta; WT, wild type.
Fig 9.
BMPR2 acts as a gatekeeper to protect ECs from activating TGFβ responses.
(A) In quiescent ECs, the major BMP delivered to ECs via the blood stream is BMP9/heterodimers of BMP9/10. The majority of BMP9 (BMP9/10) signals are transduced via ALK1/BMPR2 heteromeric receptor complexes (high affinity interactors indicated by asterisks) to induce BMP-SMAD1/5 target genes. In case of BMP6 bioavailability, as, e.g., suggested over the course of activating EC sprouting angiogenesis, BMP6 signals in conjunction with ALK2/3 and BMPR2. At the same time, bioavailability of TGFβ is supposed to be relatively low. Biologically active TGFβ signals via complexes comprising TβR2 and ALK5 to induce TGFβ-SMAD2/3 target genes. Another possible route by which TGFβ signals in ECs is via complexes comprising ALK1 (possibly also ALK2/3 when surface levels relative to ALK1 increase) allowing for phosphorylation of SMAD1/5. It was suggested that this lateral route of TGFβ signaling requires ALK5. Two possible mixed-heteromeric receptor complex formation modes are suggested: (1) the BMP and TGFβ type-1 receptors are in higher ordered neighboring oligomeric assemblies or (2) BMP and TGFβ type-1 receptors occupy TGFβ binding sites within the same complex. Eventually, these two modes do occur concomitantly in spatial and temporal proximity, e.g., as part of a clustering event. Additionally, these mixed-heteromeric assemblies are likely involved in the formation of mixed-heteromeric SMAD complexes of which little is known regarding their functionality. Upon BMPR2 deficiency, the equilibrium of these receptor complexes switches toward increased formation of TGFβ binding receptor complexes. This is because the gatekeeper function of BMPR2 is impaired. (B) BMPR2-deficient ECs gain responsiveness to TGFβ but also show increased TGFβ autostimulation when cultured under steady-state conditions. This is due to increased mechanical properties of BMPR2-deficient ECs, which co-emerge with their EndMT, the resolution of their junctional architecture, and remarkable changes in their mechanical features. As such, SMAD1 induces β1-integrin (ITGB1) expression, and ITGB1 translocates with the actomyosin scaffolding protein ILK to CCC sites, which stiffen as a result of increased stress-fiber F-actin. As a consequence of an altered matrisome expression profile, we find at CCC sites deposits of FBN1 and increased remodeling of FN. We show that latency complexes of TGFβ tether to these ECM deposits and that they can be efficiently activated by BMPR2-deficient ECs. This is facilitated extracellularly through ITGB1 binding to LAP-(TGFβ1) and intracellularly by increased ROCK-dependent actomyosin contractility acting on ILK-ITGB1. We propose that this mechanical activation of TGFβ allows continuous autostimulation for cells, which integrates into the observed increase in TGFβ responses upon BMPR2 deficiency. ALK1, activin receptor-like kinase 1; BMP, bone morphogenetic protein; BMPR2, BMP type-2 receptor; CCC, cell-to-cell contact; EC, endothelial cell; ECM, extracellular matrix; EndMT, endothelial-to-mesenchymal transition; FBN1, fibrillin-1; F-actin, filamentous actin; ILK, integrin-linked kinase; ITGB1, integrin subunit beta 1; LAP, latency-associated peptide; ROCK, Rho-associated protein kinase; SMAD, suppressor of mothers against decapentaplegic; TβR2, TGFβ type-2 receptor; TGFβ, transforming growth factor-beta.